This research identifies the socio-demographic and institutional variables associated with trainees of the SENA National Service Learning center who obtained the best results in the Icfes Saber Pro 2012 test. The data selected for this analysis correspond to the information of 9.962 participants, which 9.779 are obtained as valid observations records nationwide. The 75% of the trainees are located in the departments of Bogota D.C., Antioquia, Santander, Bolivar, Valle and Cundinamarca, while the remaining 25% is divided into 23 departments. A logit type binary choice model was used as the econometric technique to test the impact of explanatory variables on the three generic modules arranged in the tests: critical reading, quantitative reasoning and english. The probabilistic analysis concluded that the variables with the greatest positive impact on the highest levels of performance are: gender, marital status, place of residence, employment relationship, having or not having dependents, the socio-economic stratum, high school financement source (official or private), and internet connection.